R中跨数据帧的条件求和

Question

我正在尝试在 R 中复制 SUMIFS 功能。我有两个数据框。

数据框 1

allReported

ID       employeeGroup
1093     Bargaining Unit
1093     Management
1093     Non-Union
55       Bargaining Unit
55       Management
55       Non-Union

数据框 2

employeeCompSummary

ID       employeeGroup      statBenefits    regularWages
1093     Management         500.00          10000.00
1093     Management         200.00          60000.00
1093     Bargaining Unit    100.00          20000.00
1093     Bargaining Unit    150.00          30000.00
1093     Non-Union          500.00          60000.00
55       Bargaining Unit    750.00          65000.00
55       Bargaining Unit    500.00          75000.00
55       Management         250.00          45000.00
55       Management         850.00          90000.00

我正在尝试汇总 statBenefits（然后是正常工资）以创建一个新表，该表将产生以下结果：

ID       employeeGroup          statBenefits
1093     Bargaining Unit        250.00
1093     Management             700.00
1093     Non-Union              500.00
55       Bargaining Unit        1250.00
55       Management             1100.00
55       Non-Union              0.00

我尝试了以下方法：

library(data.table)
setDT(allReported)[, list(total=sum(statbenefits)), list(employeeCompSummary, employeeGroup)]

并得到以下错误：

Error in `[.data.table`(setDT(allReported), , list(total = sum(statbenefits)),  :   column or expression 1 of 'by' or 'keyby' is type list. Do not quote column names. Usage: DT[,sum(colC),by=list(colA,month(colB))]

我也试过了：

sumTest <- aggregate(allReported, by = list(employeeCompSummary), sum)

并得到以下错误：

**Error in aggregate.data.frame(allReported, by = list(employeeCompSummary),  :   arguments must have same length**

任何人都可以提供任何帮助，我们将不胜感激。我查看了其他似乎可以解决此问题但无法找到有效答案的问题。我将在多件事上完成这项任务，所以我想知道是否有任何人都知道的简单技术。与往常一样，提前感谢 Stack Overflow 上的精彩社区。​​p>

编辑两个示例表的 dput()：

allReported <- structure(list(ID = c(1093, 1093, 1093, 1093, 1093, 55, 55, 55,55), employeeGroup = c("Management", "Management", "Bargaining Unit","Bargaining Unit", "Non-Union", "Bargaining Unit", "Bargaining Unit","Management", "Management"), statBenefits = c(500, 200, 100,150, 500, 750, 500, 250, 850), regularWages = c(10000, 60000,20000, 30000, 60000, 65000, 75000, 45000, 90000)), row.names = c(NA,-9L), class = c("tbl_df", "tbl", "data.frame"))

employeeCompSummary <- structure(list(ID = c(1093, 1093, 1093, 55, 55, 55), employeeGroup =c("Bargaining Unit","Management", "Non-Union", "Bargaining Unit", "Management", "Non-Union")), row.names = c(NA, -6L), class = c("tbl_df", "tbl", "data.frame"))

 . 

Answer 1

我愿意……

library(data.table)

# don't use setDT, since who knows if it works on tibbeldies
ar = data.table(allReported)
ecs = data.table(employeeCompSummary)

ecs[, total := ar[.SD, on=.(ID, employeeGroup), sum(x.statBenefits), by=.EACHI][, V1]]

     ID   employeeGroup total
1: 1093 Bargaining Unit   250
2: 1093      Management   700
3: 1093       Non-Union   500
4:   55 Bargaining Unit  1250
5:   55      Management  1100
6:   55       Non-Union    NA

即使 OP 请求了一个新表，此代码也会将列添加到 ecs。新表和ecs 之间的行集是相同的，因此携带它们似乎是在浪费脑力。稍后删除列很简单。

如果您想知道这个“更新连接”是如何工作的，请尝试向后工作...

ar[ecs, on=.(ID, employeeGroup), sum(x.statBenefits), by=.EACHI]

# or

ar[ecs, on=.(ID, employeeGroup)]

注意原始代码中的 .SD == ecs。见?.SD。

Answer 2

您可以使用dplyr 和magrittr（对于%>%）包来执行此操作-

library(dplyr)
library(magrittr)

df1 <- structure(list(ID = c(1093, 1093, 1093, 55, 55, 55), employeeGroup =c("Bargaining Unit","Management", "Non-Union", "Bargaining Unit", "Management", "Non-Union")), row.names = c(NA, -6L), class = c("tbl_df", "tbl", "data.frame"))

df2 <- structure(list(ID = c(1093, 1093, 1093, 1093, 1093, 55, 55, 55,55), employeeGroup = c("Management", "Management", "Bargaining Unit","Bargaining Unit", "Non-Union", "Bargaining Unit", "Bargaining Unit","Management", "Management"), statBenefits = c(500, 200, 100,150, 500, 750, 500, 250, 850), regularWages = c(10000, 60000,20000, 30000, 60000, 65000, 75000, 45000, 90000)), row.names = c(NA,-9L), class = c("tbl_df", "tbl", "data.frame"))

result <- left_join(df1, df2, by = c("ID", "employeeGroup")) %>%
  group_by(ID, employeeGroup) %>%
  summarize(
    statBenefits = sum(statBenefits, na.rm = T),
    regularWages = sum(regularWages, na.rm = T)
  )
result

Answer 3

根据您的评论进行编辑：一种方法是使用 data.table 这种方式

library(data.table)
dt1 <- data.table(structure(list(ID = c(1093, 1093, 1093, 1093, 1093, 55, 55, 55,55), 
               employeeGroup = c("Management", "Management", "Bargaining Unit","Bargaining Unit", "Non-Union", "Bargaining Unit", "Bargaining Unit","Management", "Management"), statBenefits = c(500, 200, 100,150, 500, 750, 500, 250, 850), regularWages = c(10000, 60000,20000, 30000, 60000, 65000, 75000, 45000, 90000)), 
          row.names = c(NA,-9L), class = c("tbl_df", "tbl", "data.frame")), key = c("ID", "employeeGroup"))

dt2 <- data.table(structure(list(ID = c(1093, 1093, 1093, 55, 55, 55), employeeGroup =c("Bargaining Unit","Management", "Non-Union", "Bargaining Unit", "Management", "Non-Union")), 
          row.names = c(NA, -6L), class = c("tbl_df", "tbl", "data.frame")), key = c("ID", "employeeGroup"))



dt1[dt2][, lapply(.SD, sum), .SDcols = c("statBenefits", "regularWages"), by = c("ID", "employeeGroup")]

给了

ID   employeeGroup statBenefits regularWages
1:   55 Bargaining Unit         1250       140000
2:   55      Management         1100       135000
3:   55       Non-Union           NA           NA
4: 1093 Bargaining Unit          250        50000
5: 1093      Management          700        70000
6: 1093       Non-Union          500        60000

您可以稍后将 NA 值替换为 0